Bayer DSP Solutions offers a wavelet based image compression and decompression (WSI Codec) software that features both high quality image compression and low MIPS consumption. The algorithms are currently available as a compiled library for Analog Devices Blackfin DSP and PC.
A key feature of or WSI Codec is the use of 16 bit fixed point arithmetic together with the lifting scheme allowing perfect reconstruction of the original image.
WICodec is therefore also suitable for lossless image compression. Further topics are run-length encoding of wavelet coefficients, with the possibility to use various scan patterns, and entropy coding of runlengths.
The DSP implementation is complemented by a PC-based version of the codec, as typical applications require image compression by the DSP and decompression on a PC or vice versa.
The compression code was developed in a joint effort by Bayer DSP Solutions and the Department of Applied Digital Signal Processing (Prof. S. Ries) of the Southern Westphalia University of Applied Sciences (Meschede/Germany).
Datasheet

For data communication in DSP-based telephony system we have developed a full-featured HDLC transceiver that is equivalent and has been tested with standard hardware-based solutions such as Zarlink MT8952.
This HDLC algorithm implements the data link layer (Layer 2) of the OSI High-level Data Link Control protocol on TMS320C54x and C55x processors.
This solution is perfect whenever a few MIPS are available in a DSP based system and cost is an issue.
The algorithm supports multi-channel operation and can easily be integrated in a C environment. A simple API and sample C-code is provided for easy integration into the target system. The algorithm is indepedent of the target I/O configuration, i.e. it supports any type of data source and sink (serial/serial, serial/HPI etc.).
Datasheet

Universal Tone Detection and Generation

Our Universal Tone Receivers are high-performance standard functions for Analog Devices and Texas Instruments fixed-point DSP devices. They detect any tone consisting of one or two frequencies with adjustable bandwidth and sensitivity per frequency. The algorithm is targeted but not limited to PBXs and Central Office applications. The algorithm features outstanding speech immunity and very good timing resolution.
Likewise, our Universal Tone Generation and Detection algorithms are high-performance standard functions for Analog Devices and Texas Instruments fixed-point DSP devices. They can produce all kinds of signalling tones (single and dual) in PBX, Central Office or general purpose applications. The algorithms feature near perfect frequency and level precision.

Tone Detector Performance

• Detects signals or pauses >50 milliseconds least 50 milliseconds in duration. Different l Programmable center frequencies/bandwidth
• >30 dB of dynamic range
• Excellent speech immunity guaranteed by design
• Approx. 1MIPS / channel
• < 1K words program memory total
• 32 words data memory / channel
• Runs on all Analog Devices Blackfin and Texas Instruments TMS320C54x and C55x devices

Tone Generator Perfomance

• Frequency precision better than .1%
• Level precision better than .5dB
• Self Timing signal generation (automatic cadences)
• < 2MIPS / channel
• ~200 words program memory
• 6 words data memory per frequency
• Runs on all Analog Devices Blackfin and Texas Instruments TMS320C54x and C55x devices

DTMF Detection and Generation

Our DTMF receiver algorithm performs better than recommended by ETSI/ITU/Bellcore and is targeted at PBXs and Central Office applications. The algorithm features outstanding speech immunity and is superior to all traditional single chip solutions.
Likewise, our DTMF generator algorithm performs according to ETSI/CEPT targeting PBX and Central Office applications. The algorithm features near perfect frequency and level precision.

Receiver Performance

• Accepts signals and pauses of 40 msec or more
• Rejects signals and pauses of 20 msec or less
• Sustained performance in the presence of noise, frequency deviation and dial tones.
• Spans > 25 dB of dynamic range (-4dBV .. -29dBV).
• No detection at 9dB below detection level range (e.g. -38dBV for Lmin = -29dBV)
• Speech immunity guaranteed, tested with Mitel´s CM7291 tape (10 out of 30 allowed detections)
• ~ 3MIPS / channel
• ~ 2K words program memory total
• 76 words data memory / channel
• Runs on all Analog Devices Blackfin and Texas Instruments TMS320C54x and C55x devices
• Complies with CEPT TC/S 46-02, ETSI ES201235-3, Bellcore and ITU-T recommendations

Generator Performance

• Frequency precision better than .1%
• Level precision better than .5dB
• Self Timing signal generation
• Programmable Pre-emphasis (+2dB)
• ~ .2MIPS / channel
• ~ 300 words program memory
• 16 words data memory + 4 words data memory per channel
• Runs on all Analog Devices Blackfin and Texas Instruments TMS320C54x and C55x devices
• Complies with CEPT TC/S 46-02, ETSI ES201235-2, Bellcore and ITU-T recommendations

More Information...